Data governance remains a poorly understood discipline, as discussed in the post Data Governance is not Data Management. Yet, it remains the foundation of almost any data initiative, whether formally or informally,

This post discussed five programs that have a heavy dependency on effective governance.

1. Self-service Business intelligence

Self-service BI has been a goal for many organisations for more than a decade. Decision makers need information and IT (generally) is understaffed and does not have the capacity to deliver needed reports timeously,

The obvious solution is to attempt to enable decision-makers to create their own reports.

However, in spite of a spend that runs to many billions of dollars a year, many companies do not achieve the goal – trusted decision-making driven through self-service. Many self-service BI programs fizzle out, leaving decision-makers in the dark.

Understanding the disconnect between BI creators and BI consumers.

BI creators typically have an in-depth understanding of both the BI tools and the data sets available for analysis. They may have been responsible for sourcing data – so they understand its lineage and data structures – and their experience working with each data set means that they have an innate understanding of its quality.

BI consumers on the other hand do not work with data regularly. They may struggle to understand which data is the most relevant to their requirement or may have issues understanding which of several variations of the data set is the most trustworthy. They may also lose time replicating analyses or reports that already exist, due to a lack of knowledge of the environment.

These hurdles can seem insurmountable, particularly when deadlines are looming. It becomes easier to give up on the system.

How does data governance help?

A well-governed self-service environment makes information about data sets and reports easily available.

Occasional users should be able to quickly and easily compare data sets – understanding their source and quality. They should be able to understand who else has been using these data sets, and for what purpose. They should be able to see what reports already are in place for each data set, and quickly assess whether these can be reused or tweaked for a slightly different purpose.

They should be able to share their own analyses, reports and insights, in order to support the work of the broader community

Where necessary, they should be able to request access to a dataset, or report, and feel confident that their request will be processed efficiently, and without compromising data privacy and security.

Data governance, and metadata management, provide a foundation for trusted data that make it far more likely for your self-service BI program to succeed.

2. Protection of Personal Information

Data privacy is very topical right now, with PoPIA coming into full effect in just a few months.

Operationalising PoPIA requires companies to assign accountability for personal data and understand and document the valid purposes and processes for which data may be collected, processed and stored. We must understand where personal data is captured and stored, manage data access, and, in the event of a breach, assess the impact and notify affected data subjects.

This is data governance.

Data governance is about understanding who is accountable for data, setting policies for how data is used, and defining processes for dealing with data-related issues.

3. Machine learning and Artificial Intelligence

Machine learning and AI have been hot topics in tech-related media over the last year. Yet, whilst there has been a lot of hype, very few South African companies are making serious investments in machine-driven intelligence.

A lack of trust is frequently stated as a reason for delaying investment.

It is generally accepted that AI and machine learning applications are heavily dependent on the quality of the data. In this use case, data governance must look at how to define the parameters for an unbiased teaching data set and also identify how data is sourced.

Increasingly, companies must also have clarity on how the model makes decisions. This is not just an ethics problem. In various jurisdictions, the fairness of machine-driven decisions has been challenged in court.

Data governance must underpin AI and machine learning programs in order to provide oversight of models deployed in production.

4. Single customer view / Master Data Management

Often thought of as a technology or tools problem, successful master data management requires a high degree of engagement and decision-making across data silos.

Some examples of decisions that must be made:

• Defining priorities and scope for the master data solution
• Agreeing (and sourcing) critical master data elements
• Agreeing data standards and rules
• Agreeing on how to identify potential duplicates
• Agreeing on how to distribute master data so as not to lose information (source to target mappings)

Without answers to these and similar questions, it can be impossible to implement technology solutions. For example, at one client we asked “If we find the same company name, with the same company registration number, is this the same customer?” only to be told, “Not necessarily.”

Similarly, source-to-target mappings may not be obvious. For example, in one system I may have address1, address2 and address3. In another system, I may have home, work and postal addresses. It may not always be obvious how to map these.

For product/materials data I may need to agree on a global standard and ontology for product representation, or I may be required to use an internally agreed standard.

In each case, decisions made will have a significant impact on the technical implementation, tool selection and so on. Without these governance decisions, the master data project may become bogged down and never reach a visible completion.

5. Digital customer experience

The COVID-19 pandemic has accelerated the adoption of digital channels of business, both for staff who may now be working from home and for customers that are more likely to be engaging via digital channels.

This does not mean that traditional face-to-face business has disappeared. In practice, many companies are being forced to deliver omnichannel experiences – interacting with customers via the store, the call centre, social media and e-commerce sites.

Digital transformation depends on data and digital transformation drives the collection of data.

Your customer may ask a question, for example, on one channel while trying to purchase on another. If we agree that markets are conversations then a positive omnichannel customer experience relies on our being able to seamlessly follow the conversation across channels, providing the customer with the feedback that they need to complete their transaction, and to do so in an ethical manner.

Again. governance is the foundation to define how customer data will be collected, analysed, stored and shared to optimise the experience.

Conclusion

In conclusion, data governance is not just a foundation but a linchpin for success in various data-driven use cases. Its role in bridging the gap between data creators and consumers, ensuring data privacy, enabling AI and machine learning, facilitating master data management, and enhancing digital customer experiences cannot be overstated. Embracing robust data governance practices is essential for organizations aiming to harness the full potential of their data assets.

Image by Peggy und Marco Lachmann-Anke from Pixabay